Naoto Hieda

I guess the correct implementation is return t < 0.5 ? 4 * t * t * t : (t-1)*(2*t-2)*(2*t-2)+1 (from https://gist.github.com/gre/1650294) `--t` is useful but sometimes get tricky to...

inspector: 

commit:198294a works on firefox. should we close this issue...?

with `shadowBlur = 0` it works. So it seems windows+firefox doesn't like `globalCompositeOperation` and `shadowBlur` together. I don't know if it's related but quick search showed this: https://stackoverflow.com/questions/34028717/canvas-shadow-with-globalcompositeoperation-destination-out

thanks for opening this! just as a followup, I found webgl->2d copy is nevertheless slow even with `context.drawImage` (maybe because of `preserveDrawingBuffer`??) https://editor.p5js.org/micuat/sketches/eXsahOENI

thanks, it's actually funny that performs awful on firefox but fine on chrome

comparing `loadPixels`, what changed is only this line https://github.com/processing/p5.js/blob/0.9.0/src/core/p5.Renderer2D.js#L288 which makes sense - till 0.9.0, when the dirty flag is on, `loadPixels` skips the actual operation - in the example...

Thanks, now I see... I thought it's about `mag()` but in fact in the url you pointed it has `len` as a parameter of `setMag(len)`. I feel it's a bit...

that's what I thought after your first comment. The more I read the sentence I feel it sounds strange. What about > Set the magnitude of this vector to **len**...

I see. However, interestingly, when I copied the `ellipse` code to a draw function instead of calling ellipse, it works fine https://editor.p5js.org/micuat/sketches/8YZg3hoWD and manually preparing x/y/w/h works too https://editor.p5js.org/micuat/sketches/EY64BRhjl